D ISCUSSION OF THE RESULTS

To analyze the simplicity and usability, degree of overlap, relevance, completeness and effectiveness of monthly data collection sheets Heywood and Rohde’s (Heywood & Rohde 2001) work on assessment of data collection tools are used. With the purpose of investigating what the selected monthly data are used for and explore how managers in Gaborone district and in the selected health programmes use health information previous studies of similar topics have been applied. The analysis has discovered several issues in the HIS of Gaborone that correspond to the characteristics of a microsystem. This framework has in addition contributed to identify ways to improve the information flow between health facilities in Gaborone district and the selected health programmes.

7.3.1 The data collection tools

Overlapping data elements are a challenge in many poor countries (Heywood & Rohde 2001).

This is also identified as the main constraints with the data collection tools in Botswana. The only exception is the EPI programme. The overlap exists mainly due to development of additional, informal forms. Low accessibility of data from the HSU is seen as the possible explanation of this development. The managers identified a trade-off between liberating local health workers from extra work and the need for information. Too much workload on the single health worker affects the time available for the individual patient. On the other side, if information is not collected the basis for decision-making improving the health situation and the health services becomes a risk. Access to necessary information in time for decision-making is perceived as most important.

HIS in poor income countries are to a large extent influenced by external donors of programmes and international policy developers (Boerma & Stansfield 2007). The recent years’ focus on how efficient financial aids are being used has resulted in increased demand for documentation from poor countries. In the receiving countries, some side effects have been a development of similar reports, overlapping data elements and duplication of work across programmes. International policy developers appeal for equal data from all countries.

National registers of e.g. birth weight, maternal complications, vaccination coverage or psychiatric disorders are non-existent in many rich countries. Poor countries may feel

obligated to provide data to secure next years finances. Rich countries, on the other hand, are not tied up by this pressure, and are thereby free to collect the information they utilize

countries find unnecessary is highly questionable.

7.3.2 Use of the health information

An equal pattern of information use was seen among all the interviewed managers –

observing data contradicting preset goals, generating hypothesis and response by corrective actions. The pattern corresponds to Pauley et al.’s (1982) circle of information use. Higher levels seem however able to utilize information more than the lower levels. The findings are however not distinct. Examples are a clinic in Gaborone that does relate their data to possible interventions to reduce sexual transmittable infections. Contradictory, did a national planning process not take data into account before building a new facility.

The main area of information use is monitoring and evaluation of the programmes. Other areas, like planning and resource allocation seem to get less attention. This corresponds to what Otwombe (2007) identified in Kenya. The focus on the Millennium Development Goals has been an important contributor to the attention on monitoring and evaluation

internationally the past few years (Boerma & Stansfield 2007). The focus has led to improvement in producing data of better quality. Still, other areas of use are requested by researchers, for instance Boerma and Stansfield (2007).

The creation of comprehensive informal data collection tools is an indication of managers finding data necessary and useful. One of the programmes with a comprehensive informal tool does however, leave the unprocessed data “in the drawer”; contradicting the normative perspective on information use (Feldman & March 1981; Mutemwa 2006). This particular programme has only one appointed officer. Hence, the manager has limited time to process the data. Information use is thus not solely dependent upon availability of information. Here both human resources and time are influential factors.

The un-timeliness in the deliveries of reports is seen as a major constraint for information use in Gaborone. This study did not reveal any structural differences in the timeliness of the reporting between facilities as Otwombe (2007) did in Kenya. The un-timeliness is generally perceived to be connected to logistical obstacles and the capacity of the local management.

The un-timeliness is seen as an indication of poor information culture and low demand of

information by the top management. Pauley et al.’s (1982) found that regular meetings and exchange of information created a supporting environment for information use. In Gaborone, the PMTCT programme’s use of monthly meetings for report deliveries has made the

information flow transparent. In this way they succeed in creating information demand and ownership. Still, not all the facilities in Gaborone attend the meeting. One of the involved clinics had not heard about it.

7.3.3 Information flow in a microsystem perspective

Several microsystems where identified during the data analysis; single facilities, DHT, national programme units in MoH, as well as the four different health programmes

horizontally structured from the national to the facility level. The main microsystem under investigation is a system overreaching all the other microsystems, the health information system.

Poor data quality, e.g. mistyping or -calculation, the existence of both new and old forms and use of different definitions of data elements all jeopardize the use of information. In relation to timeliness of the reports from the district of Gaborone at the time of data collection, only EPI and PMTCT reports may be said to satisfy the 60 %-norm of timeliness (Hill 2007).

Currently available, validated data from HSU are four years old. Data this old are not useful in the decision-making processes as managers are not informed early enough to make proper assessments (Heywood & Rohde 2001). The case of immunization coverage rates exceeding 100 %, figure 6.4, indicates that the capacity to react upon poor data quality is low even in national level management. Thus, the data quality represents a treat to the efficiency of decisions made.

The HIS of Gaborone are highly fragmented, including many stakeholders. The flow of information is diverse across the programmes, with various ways of report deliveries. The number of computers available and the software used in data processing and analysis differ from programme to programme. Several managers identified the need for computers at district and facility level. They also request other data analysis software, like Access and SPSS. Integration of software systems across level and facilities are recognized as important by the informants. This is in accordance with the goals of the BEANISH project (2008), as computerization is recognized as a tool for improving data quality (Herbst 1999). The DHIS

and his co-authors emphasis the necessity to build capacity among information users in addition to building out technology to ensure good data quality and information use.

Computerization of all districts in Botswana was in progress during the data collection period.

According to informal sources all districts now got access to computers with DHIS software installed.

In relation to microsystem terminology, the fundamental elements of the HIS are the three levels of care; facility, district and national level. Between the different levels sequential interdependence are identified in the sense that data collection at facility level is necessary for data processing and eventually information use. One level is dependent upon the other to do their part of the job. The proximity between the different levels of responsibility regarding data collection and information use is partly indistinguishable. To determine a reciprocal relationship the interdependence should be going in both directions. As the results show, the degree of feedback to the districts and facilities are close to non-excitant. It is however seen as necessary and useful among both data providers and information users. Hence, one may argue that the interdependence between the different levels of care is also of a reciprocal character.

A high degree of coordination is thereby necessary according to Thompson (1967).

Analysis of the results revealed one “formal” and one “informal” health information system.

This correspond to what Sæbø et al. (2007) observed in Botswana. The formal system is identified by the official reports, and the delivery of the reports to the formal unit of statistics, HSU. The main coordinating mechanism associated with the formal system is direct

supervision (Mintzberg 1979), as the HSU is the national coordinating agency of health statistics in Botswana, having the main responsibility for provision of data. The informal information channels are recognized by health programmes’ development of local reports delivered elsewhere than to HSU. The informal system is related to Mintzberg’s (1979) mutual adjustment due to the informal character of the information flow. The use of informal reports has developed in a large scale in the MCH and Mental Health programme. In a managerial perspective, an informal HIS is not necessarily evil in itself. Coiera (2003:110) argues that informal information systems are useful when “data are of temporary value, of interest to very few people, are complex or the content is not predictable in advance. When tasks are infrequent, it is more cost-effective to use an informal solution”. In general it is necessary to have access to routine data over a longer period of time to compare the

population’s state of health across time and space. Health information has a great interest in several groups of the society, from policy makers, managers, and the general public. The HIS is also in general complex (Sæbø 2007; Thorseng 2008). The usefulness of an informal information system is hereby seen as limited in the case of the HIS in Botswana. Why then has the informal system aroused?

The formal system is organized according to a divisional system design, as hierarchy of information handling units, with a unit of specialists in statistics as the coordinator of the HIS.

Public health decisions are in contrast made in the single programmes. The informal system corresponds to a form of matrix system designed with a dualized authority system, as both HSU and the single programme management are giving instructions to the districts. The information channels correspond to this duality, where both the HSU and programme manager receives reports from the districts. As the HSU do not provide the needed

information on time, the informal system design has developed in order to remove problems associated with divisional designs (Scott 2003). This has happened despite Vroom’s (in Scott 2003) perception of hierarchical systems as better suitable for decision-making in cases of rapid decision-making. To access information programme managers have to communicate directly with the districts and facilities themselves. When a formal system fails, someone has to make sure the job gets done and that the population gets the services they are entitled to at the right time. The purpose of the informal system is perceived to correct the faults in the formal system. Still, it is striking that accessibility of data is low – mainly due to the un-timely deliveries of reports. In some programmes the use of information is also low, despite access to data. Hence, the problems in Botswana seem deeper embedded than the failures identified above.

In the wake of the BEANISH project there are efforts proceeding to improve the HIS in Botswana. The improvements are slowly operationalized and little work seems to have been going on to prepare the single programme for the planned changes. For instance all the programme managers had different perspective on how to improve their HIS. Furthermore, only one employer had heard about the BEANISH or DHIS. Even though improvement has been a set target area by the top management (BEANISH Project 2008; Government of Botswana 2004), little results are seen further down the organization. If no plans are done in order to secure involvement of programme managers, ownership of the improvements to come might be at risk (Pauley 1982).

The M&E officer has a key role to play when it comes to achieving good information to the management at the right time for decision to be made. The general impression is that more capacity has to be built to maximize the potential benefit of the information available at district and facility levels. National managers highly request a focal person at district level, such as the PMTCT district coordinator in Gaborone. The M&E officers function as a district HIS coordinator today.

The resources of the community are not exploited to its fullest. For the public to get access to health statistics is difficult. The major libraries, websites, or direct contact with relevant institutions do not provide the latest health statistics. When available, only brackets of the information are presented. In Uganda the inclusion of the community publishing routine health information in the newspapers facilitated information use and a means towards improving the public health situation (RHINO 2003).